Method and system for predicting marker combination of severe acute pancreatitis

Abstract

The invention relates to the technical field of marker combination prediction, and provides a method for predicting a marker combination for severe acute pancreatitis, which comprises the following steps: step 1, acquiring moderate severe / severe acute pancreatitis S, mild acute pancreatitis M and a healthy human group H peripheral blood sample; 2, obtaining a gene expression profile of each cell in each peripheral blood sample; 3, determining the cell population type and number of each peripheral blood sample, and identifying and removing low-quality cells according to quality control QC filtering data; step 4, comparing cell population types and gene RNA expressions which change remarkably in the S, M and H groups of peripheral blood samples, and screening out genes which are statistically and remarkably different in expression in the three groups of samples; and 5, carrying out actual prediction on the risks of moderate severe / severe acute pancreatitis S and mild acute pancreatitis M of the patient. The risk of severe illness can be well predicted, the clinical treatment can be timely followed, and the occurrence rate and the fatality rate of severe illness can be reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of marker combination prediction, and in particular to a marker combination method and system for predicting the severity of acute pancreatitis. Background Art

[0002] Acute pancreatitis (AP) is a disease that seriously threatens human life. Its condition is complex and changeable, and it is difficult to predict the direction of its course in the early stage of the disease. Clinically, the 2012 Atlanta grading and classification system classifies the severity of AP according to whether there is persistent organ failure, local or systemic complications. About 80% of patients have mild acute pancreatitis (MAP), 20% of patients have transient organ failure and (or) local / systemic complications that develop into moderately severe acute pancreatitis (MSAP) or persistent organ failure that develops into severe acute pancreatitis (SAP). Severe patients will have severe systemic inflammatory response syndrome and even multiple organ dysfunction syndrome, and the mortality rate can reach 30% to 40%. AP develops rapidly and its incidence rate increases year by year. Its clinical progression to severe disease is unpredictable, and persistent organ failure is the main cause of death in patients.

[0003] In recent years, a variety of clinical scoring systems have been used to predict the severity of AP, including the Ranson score, Acute physiology and chronic health evaluation II (APACHE II), Bedside index for severity in acute pancreatitis (BISAP), and modified computed tomography severity index (MCTSI). However, the above methods have limitations in early diagnosis and prediction of disease progression. Previous reports have shown that the accuracy of the Ranson score is only 82.2%, and it needs to be evaluated 48 hours after admission, which cannot achieve the role of early prediction. The APACHE II score needs to be measured in the ICU, and the APACHE II score needs to collect as many as 12 indicators, which is cumbersome in actual application and has certain limitations. The BISAP scoring system is more convenient to use than the Ranson score and APACHE II score. It calculates the total score of the five indicators within 24 hours after admission. When the total BISAP score is ≥3 points, the specificity (92%) for diagnosing SAP is good, but the sensitivity (38%) is low. Whether used alone or in combination, the clinical application in diagnosing persistent organ failure in critically ill patients is limited. The sensitivity and specificity of the best predictor within 48 hours of admission are only 75%.

[0004] Due to the acute manifestations, significant morbidity and late mortality of AP, as well as the shortcomings of traditional methods, there is an urgent need to accurately identify the risk of severe illness in the early stages of the disease and to develop a rapid and simple method that can accurately predict the risk of SAP in the early stages of AP so that timely intervention treatment can be given to reduce the mortality of SAP.

[0005] Patent application CN117265092A "Marker combination for predicting severe acute pancreatitis and its application" proposed a method for early prediction of AP using a gene model, and the markers included 10 gene models such as DUSP1, JUN, HBA2, CYBA, FOS, WDR83OS, FOSB, S100A8, JUNB, and S100A9. However, the prediction model in this patent application is relatively complex, involving a large number of genes, and the selected genes are not very interpretable from a biological perspective. The clinical operation is relatively more complicated and the cost is higher. In addition, this patent application does not provide a solution for how to obtain specific gene types in the gene model. Summary of the invention

[0006] In response to the above problems, the purpose of the present invention is to provide a method and system for predicting the severity of acute pancreatitis by combining markers, screening and compressing the genome for establishing a prediction model, retaining at least one of the most predictive genes, and the retained genes are biologically explainable, thereby improving the operability and prediction accuracy of clinical applications, helping clinicians to identify high-risk patients at an early stage and take corresponding treatment measures.

[0007] The above-mentioned object of the present invention is achieved through the following technical solutions:

[0008] A method for predicting the severity of acute pancreatitis by combining markers, comprising the following steps:

[0009] Step 1: Obtain peripheral blood samples from several patients who eventually develop moderately severe / severe acute pancreatitis S and mild acute pancreatitis M, as well as the peripheral blood samples from several healthy persons H as controls;

[0010] Step 2: performing single-cell RNA sequencing on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample;

[0011] Step 3: Importing the gene expression profile as sequencing data into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and filtering the data according to quality control QC to identify and remove low-quality cells;

[0012] Step 4: Compare the cell population types and gene RNA expressions that have changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M, and the healthy person H group, and screen out genes with statistically significant differences in expression among the three groups of samples;

[0013] Step 5: Based on the screened genes whose expressions are statistically significantly different in the three groups of samples, the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is actually predicted.

[0014] Furthermore, in step 2, single-cell RNA sequencing is performed on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample, specifically:

[0015] Decomposing the peripheral blood sample into a single cell suspension by means including enzymatic hydrolysis and mechanical separation, and filtering and counting the single cell suspension to ensure that the purity and cell concentration of the single cell suspension are appropriate;

[0016] Capturing single cells in the single cell suspension using methods including microfluidic chips and droplet technology, labeling the cells with barcodes, performing reverse transcription to convert mRNA in the cells into cDNA, amplifying the cDNA and constructing a sequencing library;

[0017] The constructed sequencing library is subjected to high-throughput sequencing using a next-generation sequencing platform including Illumina NovaSeq to obtain the gene expression profile of each cell.

[0018] Further, in step 3, the gene expression profile as sequencing data is imported into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and the data is filtered according to quality control QC to identify and remove low-quality cells, specifically:

[0019] Importing the gene expression profile as the sequencing data into Seurat software, wherein the sequencing data exists in a matrix form, with genes as rows and cells as columns;

[0020] Cells with less than 500 genes and genes expressed in no more than 5 cells were filtered out based on quality control QC;

[0021] Normalizing the sequencing data using the LogNormalize method to adjust the sequencing depth difference, and scaling the normalized sequencing data so that the gene expression data has a distribution with a mean of 0 and a variance of 1;

[0022] Use PCA to reduce the dimension of the normalized data, extract the main variation information, select the first few principal components for cluster analysis, and use clustering algorithms including Louvain or Leiden to cluster cells, grouping cells with similar gene expression patterns, with each group representing a cell population;

[0023] Based on known marker genes, each of the cell populations is annotated, the cell population type of each of the cell populations is determined, the number of cells in each of the cell populations is counted, and the proportion of each of the cell population types in the peripheral blood sample is determined.

[0024] Further, in step 4, the cell population types and gene RNA expressions that changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M and the healthy person H group were compared, and genes with statistically significant differences in expression in the three groups of samples were screened out, specifically:

[0025] Perform differential expression analysis on all genes to obtain the p-value of each gene, sort all genes in order from small to large p-value, and select the top 100 genes with the smallest p-value, which are the genes with the most significant differential expression in statistics;

[0026] The expression data of the 100 genes in the healthy person H group, the moderate severe / severe acute pancreatitis S group, and the mild acute pancreatitis M group were extracted, and the genes were screened according to the change trend of the expression data in the three groups, and the screening criterion was to screen out the genes whose gene expression levels increased or decreased in the order of the healthy person H group, the mild acute pancreatitis M group, and the moderate severe / severe acute pancreatitis S group;

[0027] For the screened genes, the average expression level in each of the peripheral blood samples was calculated, and after a round of minimum expression level screening, genes with an average expression level of not less than 2 TPM in each of the peripheral blood samples were retained.

[0028] Furthermore, after step 4, the method further includes: using the average expression of the gene screened in step 4 in each of the peripheral blood samples as an independent variable and the sample group of each of the peripheral blood samples as a dependent variable to establish a lasso model for training, wherein the lasso model automatically compresses the coefficients of the gene and retains at least one gene that is most important for predicting the dependent variable, specifically:

[0029] Prepare the data:

[0030] Independent variable X: the average expression level of the gene screened in step 4 in each of the peripheral blood samples;

[0031] Dependent variable: sample group of each peripheral blood sample, wherein the healthy person H is 1, the mild acute pancreatitis M is 2, and the moderately severe / severe acute pancreatitis S is 3;

[0032] Normalize data:

[0033] Before performing the lasso model regression, the independent variables were standardized to ensure that the expression levels of different genes were in the same dimension;

[0034] Build the lasso model:

[0035] Using the lasso model for training, the lasso model regression is optimized by minimizing the following objective function:

[0036]

[0037] Among them, λ is the regularization parameter, which controls the strength of the penalty term, and βj is the regression coefficient of the independent variable, y j is the actual group, is the predicted outcome variable, n is the number of peripheral blood samples, and p is the number of independent variables;

[0038] Choose the best parameters:

[0039] The optimal regularization parameter λ is selected through cross-validation to balance the complexity and fit of the model;

[0040] Identify the most predictive genes:

[0041] The lasso model automatically compresses the coefficients of genes with low predictive power to zero, thereby selecting at least one gene that is most important for predicting the dependent variable.

[0042] Further, in step 5, based on the genes screened out with statistically significantly different expressions in the three groups of samples, the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is actually predicted, specifically:

[0043] Obtaining the peripheral blood sample of the patient to be predicted in clinic, and collecting cells in the peripheral blood sample;

[0044] Extracting the genes for prediction screened in step 4 and the corresponding RNA expression of the genes from the cells of the peripheral blood sample, and determining the expression level of each gene;

[0045] The prediction formula of the lasso model is used for prediction, and the formula is as follows:

[0046]

[0047] in, is the predicted outcome variable, β0 is the intercept of the lasso model, and β j is the regression coefficient of the ith gene, x i is the expression level of the ith gene, and is the number of genes screened;

[0048] A threshold value of the predicted outcome variable for the moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is set, and the predicted value is classified according to the threshold value.

[0049] The method for predicting the combination of markers for the severity of acute pancreatitis also includes: replacing the lasso model with models including support vector machine SVM, logistic regression, and random forest to obtain at least one gene that is most important for predicting the dependent variable.

[0050] A system for predicting the severity of acute pancreatitis using a combination of markers for executing the above-mentioned method for predicting the severity of acute pancreatitis using a combination of markers, comprising:

[0051] A sample acquisition module, used to obtain peripheral blood samples from several patients who eventually develop moderately severe / severe acute pancreatitis S and mild acute pancreatitis M, as well as the peripheral blood samples from several healthy persons H as controls;

[0052] A cell sequencing module, used to perform single-cell RNA sequencing on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample;

[0053] A cell population classification module, for importing the gene expression profile as sequencing data into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and filtering the data according to quality control QC to identify and remove low-quality cells;

[0054] A gene screening module, used to compare the cell population types and gene RNA expressions that have changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M, and the healthy person H group, and screen out genes with statistically significant differences in expression in the three groups of samples;

[0055] The pancreatitis prediction module is used to actually predict the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M based on the screened genes whose expressions are statistically significantly different in the three groups of samples.

[0056] A computer device comprises a memory and one or more processors, wherein the memory stores computer codes, and when the computer codes are executed by the one or more processors, the one or more processors execute the above method.

[0057] A computer-readable storage medium stores computer codes. When the computer codes are executed, the above method is executed.

[0058] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0059] (1) The combination of markers for predicting severe pancreatitis screened by the present invention can well predict the risk of severe illness, help clinicians to follow up on targeted treatment in a timely manner, reduce the incidence and mortality of severe illness, and save more patients. At the same time, the screened genes can be used as biomarkers to find key SAP cell populations, help to deeply understand the mechanism of SAP occurrence, and find key factors to block the occurrence of SAP. In addition, this model only retains a small number of genes, has strong interpretability, low clinical testing costs, and fast and reliable results, bringing more social and economic benefits to hospitals and patients.

[0060] (2) Single-cell sequencing technology was used to screen gene combinations significantly associated with SAP, and screening was performed based on the expression trend to ensure that the genes had biological significance and clinical application value. A simplified prediction model based on the expression of a small number of genes was established, which has rapid and accurate detection capabilities and can complete the determination of the expression level of relevant genes in patients within 90 minutes. The diagnosis of severe acute pancreatitis significantly improved the accuracy and convenience of early prediction of SAP. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 The overall flow chart of the method for predicting the severity of acute pancreatitis by using a combination of markers according to the present invention;

[0062] Figure 2 The present invention is based on the screened genes whose expression is statistically significantly different in the three groups of samples to actually predict the risk of pancreatitis;

[0063] Figure 3 This is an overall structural diagram of the marker combination system for predicting the severity of acute pancreatitis according to the present invention. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0065] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0066] First embodiment

[0067] like Figure 1 As shown, this embodiment provides a method for predicting the severity of acute pancreatitis by a combination of markers, comprising the following steps:

[0068] Step 1: Obtain peripheral blood samples from several patients who eventually develop moderately severe / severe acute pancreatitis S and mild acute pancreatitis M, as well as the peripheral blood samples from several healthy persons H as controls.

[0069] Specifically, before calculating the combination of markers for predicting severe pancreatitis, it is necessary to obtain as many peripheral blood samples as possible from patients with moderately severe / severe acute pancreatitis S and mild acute pancreatitis M, and it is also necessary to obtain peripheral blood samples from healthy controls H as controls. For example, in this embodiment, peripheral blood samples of 13 patients with acute pancreatitis collected on the first day of hospitalization and peripheral blood samples of 2 healthy controls were obtained. Among the peripheral blood samples of 13 patients with acute pancreatitis, 7 eventually developed into moderately severe / severe acute pancreatitis S, and 6 developed into mild acute pancreatitis M.

[0070] Step 2: Perform single-cell RNA sequencing on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample.

[0071] Specifically, the implementation of step 2 may include the following process:

[0072] Decomposing the peripheral blood sample into a single cell suspension by means including enzymatic hydrolysis and mechanical separation, and filtering and counting the single cell suspension to ensure that the purity and cell concentration of the single cell suspension are appropriate;

[0073] Capturing single cells in the single cell suspension using methods including microfluidic chips and droplet technology, labeling the cells with barcodes, performing reverse transcription to convert mRNA in the cells into cDNA, amplifying the cDNA and constructing a sequencing library;

[0074] The constructed sequencing library is subjected to high-throughput sequencing using a next-generation sequencing platform including Illumina NovaSeq to obtain the gene expression profile of each cell.

[0075] In this example, single-cell RNA sequencing was performed on each of the 15 peripheral blood samples obtained to obtain the gene expression profile of each cell in each peripheral blood sample.

[0076] Step 3: Import the gene expression profile as sequencing data into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and filter the data according to quality control QC to identify and remove low-quality cells.

[0077] Specifically, the implementation of step 3 may include the following process:

[0078] Importing the gene expression profile as the sequencing data into Seurat software, wherein the sequencing data exists in a matrix form, with genes as rows and cells as columns;

[0079] Cells with less than 500 genes and genes expressed in no more than 5 cells were filtered out based on quality control QC;

[0080] Normalizing the sequencing data using the LogNormalize method to adjust the sequencing depth difference, and scaling the normalized sequencing data so that the gene expression data has a distribution with a mean of 0 and a variance of 1;

[0081] Use PCA to reduce the dimension of the normalized data, extract the main variation information, select the first few principal components for cluster analysis, and use clustering algorithms including Louvain or Leiden to cluster cells, grouping cells with similar gene expression patterns, with each group representing a cell population;

[0082] Based on known marker genes, each of the cell populations is annotated, the cell population type of each of the cell populations is determined, the number of cells in each of the cell populations is counted, and the proportion of each of the cell population types in the peripheral blood sample is determined.

[0083] Step 4: Compare the cell population types and gene RNA expressions that have changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M and the healthy person H group, and screen out genes with statistically significant differences in expression among the three groups of samples.

[0084] Specifically, the implementation of step 4 may include the following process:

[0085] Since the number of peripheral blood samples is sometimes relatively small, for example, there are only 15 samples in this embodiment. The adjusted p-value after multiple verification cannot find significantly differentially expressed genes. Therefore, we perform differential expression analysis on all genes, obtain the p-value of each gene, sort all genes in order from small to large p-value, and select the top 100 genes with the smallest p-value, which are statistically the most significantly differentially expressed genes;

[0086] The expression data of these 100 genes in the healthy person H group, the moderate severe / severe acute pancreatitis S group, and the mild acute pancreatitis M group were extracted, and the expression data were screened according to the change trend of the three groups. The screening criteria were to screen out the genes whose gene expression levels increased or decreased in the order of the healthy person H group, the mild acute pancreatitis M group, and the moderate severe / severe acute pancreatitis S group. The basis for this screening is that from a biological point of view, the three groups of samples of normal control H, mild acute pancreatitis M, and moderate severe / severe acute pancreatitis S are in a progressive relationship, so the differential expression of genes that can truly reflect the different groups must also follow an increasing or decreasing trend, otherwise it will conflict with biological theory.

[0087] For example, suppose the expression levels of genes A, B, C, and D in the three groups are:

[0088] Gene A:

[0089] Healthy people group H: 10

[0090] Mild acute pancreatitis M group: 20

[0091] Moderately severe / severe acute pancreatitis S group: 30

[0092] The expression level increased and met the screening criteria.

[0093] Gene B:

[0094] Healthy people group H: 50

[0095] Mild acute pancreatitis M group: 30

[0096] Moderately severe / severe acute pancreatitis S group: 10

[0097] The expression level increased and met the screening criteria.

[0098] Gene C:

[0099] Healthy people group H: 30

[0100] Mild acute pancreatitis M group: 50

[0101] Moderately severe / severe acute pancreatitis S group: 10

[0102] The expression level neither increases nor decreases, and does not meet the screening criteria.

[0103] Gene D:

[0104] Healthy people group H: 40

[0105] Mild acute pancreatitis M group: 30

[0106] Moderately severe / severe acute pancreatitis S group: 50

[0107] The expression level neither increases nor decreases, and does not meet the screening criteria.

[0108] Therefore, genes A and B sign the screening criteria.

[0109] For the genes screened out, the average expression (TPM, Transcripts Per Million) in each of the peripheral blood samples was calculated, and a round of minimum expression screening was performed, and the genes whose average expression in each of the peripheral blood samples was not less than 2TPM were retained. The basis for this screening is that as an effective biomarker, it should have a higher expression in all sample groups, so that the results are more stable and accurate when comparing the expression differences between groups. Otherwise, if the expression in a certain group is very low, it is easy to cause a very high change multiple when comparing the expression with other groups, resulting in unstable results. In addition, it is not easy to accurately measure the actual expression when the expression is too low during sequencing, which will increase the instability of the results.

[0110] After step 4, the method further includes taking the average expression of the gene screened in step 4 in each peripheral blood sample as an independent variable and the sample group of each peripheral blood sample as a dependent variable to establish a lasso model for training, wherein the lasso model automatically compresses the coefficients of the gene and retains at least one gene that is most important for predicting the dependent variable, specifically:

[0111] Prepare the data:

[0112] Independent variable X: the average expression level of the gene screened in step S4 in each of the peripheral blood samples;

[0113] Dependent variable: sample group of each peripheral blood sample, wherein the healthy person H is 1, the mild acute pancreatitis M is 2, and the moderately severe / severe acute pancreatitis S is 3;

[0114] Normalize data:

[0115] Before performing the lasso model regression, the independent variables were standardized to ensure that the expression levels of different genes were in the same dimension;

[0116] Build the lasso model:

[0117] Using the lasso model for training, the lasso model regression is optimized by minimizing the following objective function:

[0118]

[0119] Among them, λ is the regularization parameter, which controls the strength of the penalty term, and βj is the regression coefficient of the independent variable, y j is the actual group, is the predicted outcome variable, n is the number of peripheral blood samples, and p is the number of independent variables;

[0120] Choose the best parameters:

[0121] The optimal regularization parameter λ is selected through cross-validation to balance the complexity and fit of the model;

[0122] Identify the most predictive genes:

[0123] The lasso model automatically compresses the coefficients of genes with low predictive power to zero, thereby selecting at least one gene that is most important for predicting the dependent variable.

[0124] For example, in this embodiment, data from 15 peripheral blood samples were used. After screening in step S4, the average expression of the top 10 genes in each peripheral blood sample was retained as the independent variable, and the group of each peripheral blood sample was used as the dependent variable (normal control H is 1, M is 2, S is 3), and the lasso model was established. The lasso model can select independent variables by itself, and finally determined the 4 most predictive genes. Appendix 1 is a 4-gene model.

[0125] Table 1

[0126]

[0127] Finally, CMTM2, S100A8, CST7 and IL32 were selected as genes with predictive ability selected by the LASSO model.

[0128] It should be noted that the above examples in this embodiment are only examples. In actual prediction, step S5 is not necessary, and the 10 genes selected in step 4 can be directly retained for prediction. At the same time, the genes further compressed in step S5 are not necessarily 4, and can also be any one or more of the 4 genes selected. This is not limited in this embodiment.

[0129] Preferably, in this embodiment, the four genes predicted by the LASSO model in step 4 are used as genes predicted by the model.

[0130] Step 5: Based on the screened genes whose expressions are statistically significantly different in the three groups of samples, the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is actually predicted.

[0131] like Figure 2 As shown, in this embodiment, step 5 is specifically as follows:

[0132] Obtaining the peripheral blood sample of the patient to be predicted in clinic, and collecting cells in the peripheral blood sample;

[0133] Extracting the genes for prediction screened in step 4 and the corresponding RNA expression of the genes from the cells of the peripheral blood sample, and determining the expression level of each gene;

[0134] The prediction formula of the lasso model is used for prediction, and the formula is as follows:

[0135]

[0136] in, is the predicted outcome variable, β0 is the intercept of the lasso model, and β j is the regression coefficient of the ith gene, x i is the expression level of the ith gene, and is the number of genes screened;

[0137] A threshold value of the predicted outcome variable for the moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is set, and the predicted value is classified according to the threshold value.

[0138] In addition, in this embodiment, models including support vector machine SVM, logistic regression, and random forest can be used to replace the lasso model to obtain at least one gene that is most important for predicting the dependent variable.

[0139] Second embodiment

[0140] This embodiment verifies the model based on the predicted value of the first embodiment, specifically including:

[0141] In order to test the performance of the model, 50 bootstrap samplings were performed on 13 samples (excluding 2 healthy samples), and their performance was calculated respectively. The results showed that the prediction accuracy of the model for S (positive) / M (negative) reached an average of 88%, and the average AUROC was 90.8, which has excellent performance and can help clinicians confirm the risk of patients transforming into moderately severe / severe acute pancreatitis S. As shown in Table 2.

[0142] Table 2

[0143]

[0144]

[0145]

[0146] Second embodiment

[0147] like Figure 3 As shown, this embodiment provides a system for predicting the severity of acute pancreatitis using a combination of markers for executing the method for predicting the severity of acute pancreatitis using a combination of markers in the first embodiment, comprising:

[0148] The sample acquisition module 1 is used to obtain peripheral blood samples from several patients who eventually develop moderately severe / severe acute pancreatitis S and mild acute pancreatitis M, as well as the peripheral blood samples from several healthy persons H as controls;

[0149] Cell sequencing module 2, used for performing single-cell RNA sequencing on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample;

[0150] A cell population classification module 3 is used to import the gene expression profile as sequencing data into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and filter the data according to quality control QC to identify and remove low-quality cells;

[0151] Gene screening module 4, used to compare the cell population types and gene RNA expressions that have changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M and the healthy person H group, and screen out genes with statistically significant differences in expression in the three groups of samples;

[0152] The pancreatitis prediction module 5 is used to actually predict the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M based on the screened genes with statistically significantly different expressions in the three groups of samples.

[0153] A computer-readable storage medium stores a computer code. When the computer code is executed, the above method is executed. A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program. The program can be stored in a computer-readable storage medium. The storage medium can include: a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc.

[0154] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

[0155] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0156] It should be noted that the above embodiments can be freely combined as needed. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.

Claims

1. A method for predicting the severity of acute pancreatitis by combining markers, characterized in that: The following steps are involved: Step 1: Obtain peripheral blood samples from several patients who eventually develop moderately severe / severe acute pancreatitis S and mild acute pancreatitis M, as well as the peripheral blood samples from several healthy persons H as controls; Step 2: performing single-cell RNA sequencing on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample; Step 3: Importing the gene expression profile as sequencing data into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and filtering the data according to quality control QC to identify and remove low-quality cells; Step 4: Compare the cell population types and gene RNA expressions that have changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M, and the healthy person H group, and screen out genes with statistically significant differences in expression among the three groups of samples; Step 5: Based on the screened genes whose expressions are statistically significantly different in the three groups of samples, the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is actually predicted.

2. The method for predicting the severity of acute pancreatitis according to claim 1, characterized in that: In step 2, single-cell RNA sequencing is performed on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample, specifically: Decomposing the peripheral blood sample into a single cell suspension by means including enzymatic hydrolysis and mechanical separation, and filtering and counting the single cell suspension to ensure that the purity and cell concentration of the single cell suspension are appropriate; Capturing single cells in the single cell suspension using methods including microfluidic chips and droplet technology, labeling the cells with barcodes, performing reverse transcription to convert mRNA in the cells into cDNA, amplifying the cDNA and constructing a sequencing library; The constructed sequencing library is subjected to high-throughput sequencing using a next-generation sequencing platform including Illumina NovaSeq to obtain the gene expression profile of each cell.

3. The method for predicting the severity of acute pancreatitis according to claim 1, characterized in that: In step 3, the gene expression profile as sequencing data is imported into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and the data is filtered according to quality control QC to identify and remove low-quality cells, specifically: Importing the gene expression profile as the sequencing data into Seurat software, wherein the sequencing data exists in a matrix form, with genes as rows and cells as columns; Cells with less than 500 genes and genes expressed in no more than 5 cells were filtered out based on quality control QC; Normalizing the sequencing data using the LogNormalize method to adjust the sequencing depth difference, and scaling the normalized sequencing data so that the gene expression data has a distribution with a mean of 0 and a variance of 1; Use PCA to reduce the dimension of the normalized data, extract the main variation information, select the first few principal components for cluster analysis, and use clustering algorithms including Louvain or Leiden to cluster cells, grouping cells with similar gene expression patterns, with each group representing a cell population; Based on known marker genes, each of the cell populations is annotated, the cell population type of each of the cell populations is determined, the number of cells in each of the cell populations is counted, and the proportion of each of the cell population types in the peripheral blood sample is determined.

4. The method for predicting the severity of acute pancreatitis according to claim 1, characterized in that: In step 4, the cell population types and gene RNA expressions that changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M and the healthy person H group were compared, and genes with statistically significant differences in expression in the three groups of samples were screened, specifically: Perform differential expression analysis on all genes to obtain the p-value of each gene, sort all genes in order from small to large p-value, and select the top 100 genes with the smallest p-value, which are the genes with the most significant differential expression in statistics; The expression data of the 100 genes in the healthy person H group, the moderate severe / severe acute pancreatitis S group, and the mild acute pancreatitis M group were extracted, and the genes were screened according to the change trend of the expression data in the three groups, and the screening criterion was to screen out the genes whose gene expression levels increased or decreased in the order of the healthy person H group, the mild acute pancreatitis M group, and the moderate severe / severe acute pancreatitis S group; For the screened genes, the average expression level in each peripheral blood sample was calculated, and then a round of minimum expression level screening was performed to retain the genes whose average expression level in each peripheral blood sample was not less than 2 TPM.

5. The method for predicting the severity of acute pancreatitis according to claim 4, characterized in that: After step 4, the method further includes: using the average expression of the gene screened in step 4 in each peripheral blood sample as an independent variable and the sample group of each peripheral blood sample as a dependent variable to establish a lasso model for training, wherein the lasso model automatically compresses the coefficients of the gene and retains at least one gene that is most important for predicting the dependent variable, specifically: Prepare the data: Independent variable X: the average expression level of the gene screened in step 4 in each of the peripheral blood samples; Dependent variable: sample group of each peripheral blood sample, wherein the healthy person H is 1, the mild acute pancreatitis M is 2, and the moderately severe / severe acute pancreatitis S is 3; Normalize data: Before performing the lasso model regression, the independent variables were standardized to ensure that the expression levels of different genes were in the same dimension; Build the lasso model: Using the lasso model for training, the lasso model regression is optimized by minimizing the following objective function: Among them, λ is the regularization parameter, which controls the strength of the penalty term, and β j is the regression coefficient of the independent variable, y j is the actual group, is the predicted outcome variable, n is the number of peripheral blood samples, and p is the number of independent variables; Choose the best parameters: The optimal regularization parameter λ is selected through cross-validation to balance the complexity and fit of the model; Identify the most predictive genes: The lasso model automatically compresses the coefficients of genes with low predictive power to zero, thereby selecting at least one gene that is most important for predicting the dependent variable.

6. The method for predicting the severity of acute pancreatitis according to claim 5, characterized in that: In step 5, based on the genes screened out with statistically significantly different expressions in the three groups of samples, the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is actually predicted, specifically: Obtaining the peripheral blood sample of the patient to be predicted in clinic, and collecting cells in the peripheral blood sample; Extracting the genes for prediction screened in step 4 and the corresponding RNA expression of the genes from the cells of the peripheral blood sample, and determining the expression level of each gene; The prediction formula of the lasso model is used for prediction, and the formula is as follows: in, is the predicted outcome variable, β0 is the intercept of the lasso model, and β j is the regression coefficient of the ith gene, x i is the expression level of the ith gene, and is the number of genes screened; A threshold value of the predicted outcome variable for the moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M is set, and the predicted value is classified according to the threshold value.

7. The method for predicting the severity of acute pancreatitis according to claim 1, characterized in that: Also includes: Models including support vector machine (SVM), logistic regression and random forest are used to replace the lasso model to obtain at least one gene that is most important for predicting the dependent variable.

8. A system for predicting the severity of acute pancreatitis using a combination of markers for executing the method for predicting the severity of acute pancreatitis using a combination of markers as described in any one of claims 1 to 7, characterized in that: include: A sample acquisition module, used to obtain peripheral blood samples from several patients who eventually develop moderately severe / severe acute pancreatitis S and mild acute pancreatitis M, as well as the peripheral blood samples from several healthy persons H as controls; A cell sequencing module, used to perform single-cell RNA sequencing on all the peripheral blood samples to obtain the gene expression profile of each cell in each peripheral blood sample; A cell population classification module, for importing the gene expression profile as sequencing data into Seurat software for analysis and clustering to determine the cell population type and quantity of each peripheral blood sample, and filtering the data according to quality control QC to identify and remove low-quality cells; A gene screening module, used to compare the cell population types and gene RNA expressions that have changed significantly in the peripheral blood samples of the moderately severe / severe acute pancreatitis S, the mild acute pancreatitis M, and the healthy person H group, and screen out genes with statistically significant differences in expression in the three groups of samples; The pancreatitis prediction module is used to actually predict the risk of the patient's moderately severe / severe acute pancreatitis S and the mild acute pancreatitis M based on the screened genes whose expressions are statistically significantly different in the three groups of samples.

9. A computer device comprising a memory and one or more processors, wherein the memory stores computer codes, and when the computer codes are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 7. 10 . A computer-readable storage medium storing a computer code. When the computer code is executed, the method according to claim 1 is executed.

Images